experiment on high capacity backhaul transmission link ...transmission link aggregation solution for...

Experiment on High Capacity Backhaul Transmission Link Aggregation Solution for 5G

Networks

Mohammad Reza KouchakiWireless Network Planning and Optimization

Mobinnet Telecom

Tehran, Iran

[email protected]

Mohammad DabibiSchool of Electrical and Computer Eng.

The University of Tehran

Tehran, Iran

[email protected]

Abstract—Nowadays, commercial 5G networks are starting togrow up rapidly. Requirements such as 1-10Gbps connections toendpoints, one millisecond end-to-end round trip delay - latency,1000x bandwidth per unit area, 10-100x number of connecteddevices, 99.999% availability, 100% coverage, 90% reduction innetwork energy usage also extended compare to 4G networks.Existing microwave link bandwidths can no longer support ever-escalating requirements for 5G Backhaul capacity due to expo-nentially growing demand for mobile traffic by the arrival of thenew portable gadgets. Spectrum resources of typical frequencybands (6 to 42 GHz) are proving insufficiency for providinghigh microwave backhaul capacity, making network backhaulcapacity expansion increasingly more difficult. Face with thechallenges, This paper presents an experiment on 5G BackhaulSmall-Cell, includes an advanced Microwave Link Aggregation(MLA) technique to provide enough Backhaul capacities for5G network. In this experiment, Mentum Ellipse used as amicrowave planning tool for analytical simulation for proposedlink aggregation, which offers 6.27 Gbps. The pilot implementedin the dense urban area delivers 6.073 Gbps in the proposedCross Dual Band Link Aggregation (CDBLA) solution. CDBLAenables the integration of Common-Band microwave (CBM) (6to 42 GHz) and E-Band microwave (EBM) (71 to 76 and 81 to86 GHz) into one high capacity link to support large bandwidthand long-distance transmission.

I. INTRODUCTION

5G is the next-generation cellular network that is expectedto be highly efficient and fast, as well as being able to supportmore users, billions of smart devices, more services, andnew use cases without a corresponding impact on the cost.Research and standardization work has been addressing fromradio spectrum allocation, densification, massive multiple-input-multiple-output (MIMO) antenna, carrier aggregation(CA), inter-cell interference mitigation techniques, andcoordinated multi-point processing[1], [2], [3]. In Addition, anew network bottleneck has emerged: the Backhaul networkwhich will allow to interconnect and support billions ofdevices from the core network[4]. Up to current 4G cellularnetworks, the significant challenges to meet the Backhaulrequirements were capacity, availability, deployment cost, andlong-distance reach[4]. However, as 5G network capabilities

and services added to 4G cellular network, the Backhaulnetwork would face additional ultra-interconnection natureof the network challenge, resulting in Microwave Backhaulnetwork need to maintain multi-Giga traffic from the networkhubs to today’s cellular networks that are infeasible to meetthis requirement in terms of capacity, availability, latency, andcost-efficiency.Previous studies achieved high data rates by combining newtechnologies such as small cell, millimetre wave (mmWave),full-duplex, beamforming (BF) and MIMO networks[5],[6]. The previous experiment in [7] has transmitted 1Gbpsthroughput over the frequency range without mobility over200 meters in non-line of sight and 1.7 Km in line ofsight (LOS) mmWave technology at 27.925 GHz with 4×8planar array prototype to provides acceptable performance.Regarding [3], the mmWave radio frequency signal is moresensitive to path losses such as building and environmentblockage. Therefore, it presents a shorter distance betweenbase stations and users, which yields the advent of smallcells[7], [8]. Furthermore, in [9] that 5G base station usedmassive MIMO, the distance between small cells and usersdecreased more. As a result, in this paper we experimentmicrowave interface with aggregation techniques betweenradio base stations and 5G small cells to eliminate shortdistance and low throughput issue.This paper introduces the advanced microwave linkaggregation solution between 5G small-cells and radiobase stations compare to the previous works. In section II,remarkable previous approaches and scenarios are studiedand analyzed for 5G Backhaul implementation, then ouradvanced scenario and CDBLA model is introduced. In partIII, the proposed microwave link aggregation is designed andanalyzed in a software tool. In the next section, the proposedsolution is implemented and measured in a practical location.Finally, the conclusion results and comparison with classicalsolutions presented in section V.

II. METHODOLOGY

A. 5G BACKHAUL TOPOLOGY

Mobile and fixed broadband networks are continuallyevolving, leading to an exponential increase in the Backhaul

______________________________________________________PROCEEDING OF THE 26TH CONFERENCE OF FRUCT ASSOCIATION

ISSN 2305-7254

capacity of base stations. And as 5G fast approaches, themicrowave is facing the growing challenges. Common-bandmicrowave provides highly reliable transmission line. Thereare some techniques for Common-band microwave throughputenhancement such as carrier aggregation (CA), MIMO dual-band with V/H polarization and co-channel interference can-cellation (CCIC) are discussed in [10], [11], [12]. Althoughthese techniques increase throughput, its spectrum resourcesare still insufficient, and maximum throughput is (< 2Gbit/s),making capacity expansion challenging to raise more.E-band microwave supports up to 4.5 Gbit/s, which satisfies thethroughput requirements of 5G. However, its reliability reducesby distance and rain fades. Moreover, frequency channelinterference of adjacent links prevents using its full bandwidth(BW) capacity, making it incapable of meeting the medium-distance Backhaul requirements of small-cell stations[12]. So,we planned and experiment with a solution that combines thelong-distance and high-reliability capabilities of common-bandlinks with the high throughput of E-band links. Fig.1 showsour experiment small cell Backhaul network topology that basestations use fibre interface and proposed CDBLA to connectwith core and small cells, respectively.

Fiber backhaul

Core Network

Campus Network

5GSmall-Cell

Campus building

Fig. 1. 5G small cell Backhaul topology

B. PROPOSED MODEL

Analyzing environmental conditions such as fog or rainare particularly important when using super or extremelyhigh frequencies (30-300 GHz)[13]. Losses incurred due toatmospheric or rain attenuation tend to increase at the mmWavewavelengths. Atmospheric attenuation is based on the ITU-R P676 that includes pressure, temperature and water con-centration [14]. These parameters can be defined based onthe location of the base station and the ITU classification byregion. An approximate attenuation Arain due to rainfall inthis scenario is:

Arain = γrainde (1)

Where de is the effective length, the corrected value of thereal path length, and also, it is the function of frequency andthe actual climatic zone that formulated in (2):

de =d

1 + d/d0

(2)

With d in kilometres and d0 = 35e−0.5R. Also,γrain(dB/Km) is based on coefficient α and β calculation forthe rainfall intensity R(mm/h) that is defined:

γrain = α(R)β (3)

In our experiment the implemented path length is 3.2km,therefore de = 3km. To estimate the long-term statistics of rainattenuation, the recommendation ITU-R-P530[15] can be ap-plied for E-band as well[16]. The validity of (1) is fair enoughif we consider long (12 months) measurement period withTableI value for α and β for frequency dependent coefficientsfor estimation specific rain attenuation for horizontal (H) andvertical (V) polarization.

TABLE I. REGRESSION COEFFICIENTS OF ITU-R P.838-1

Frequency(GHz) βH βV αH αV

18 0.0751 0.0691 1.099 1.065

80 0.975 0.906 0.769 0.769

Rain attenuation at specific mmWave frequencies impactsthe accuracy of predictions. Rain attenuation for differentrainfall rate R can be found in ITU-R recommendation [14],[15] as shown in Fig.2. The rain attenuation is very significantin mm-waves. For example, rainfall rates of medium (12.5mm/h) and downpour (50 mm/h) yield to about 7 and 20dB/km attenuations in E-band at 70∼80 GHz. For the caseof severe (tropical) rainfalls (100 mm/h), the attenuation canreach up to 30 dB/km. However, this regularly occurs onlyin short bursts and is associated with a critical weather eventthat moves quickly across the link path[17]. Therefore, therainfall rate is always associated with time probability or linkavailability of the average year.

Frequency (GHz)

Rain

Atte

nuat

ion

(dB

/Km

)

E-band

18GH

z

2.5mm/h

12.5mm/h

50mm/h

100mm/h

Fig. 2. Rain attenuation for different rainfall based on [18]. (Frequency 18Gand E-band are specified for proposed CDBLA implementation.)

In general, Availability calculates as:

Availability[%] =100(total usage time− downtime)

total usage time


---------------------------------------------------------------------------- 226 ----------------------------------------------------------------------------

Following ITU-R, the highest link availability is 99.999%.Thus, according to common Service Level Agreement (SLA)between a service provider and its internal or external cus-tomers the depicted in Fig.3; link availabilities between99.99% and 99.999% or 0.01 to 0.001% (26.17 and 52.6 min)of yearly times are considered for the common-radio links andlink availabilities between 99.9% and 99.99% or 0.1 to 0.01%(1hours and 8hours) of annual times are considered for theE-band radio links in this work. In this work, according to thecross-link aggregation in section I, we aim to reach average99.99% in medium and downpour rainfalls environment inITU-R P.530 7-16[15] rain model at the experimental 3.2Kmlink distance for customers with high priority services[19].

Low-priorityServices

99.9% 8.76hrsE-Band

High-priorityServices

99.99% 52.6minCDBLA

Voice 99.999% 5.3minCommon Band

SLA Availability/Annual down Time

Fig. 3. Validity of Network Performance

In order to ensure network performance is optimal, wevalidated hardware settings and network design by availabilityand Quality of Service performance (QoS). In our work, High-priority services transmitted over common-band microwaveto achieve carrier-class availability and Low-priority servicestransmitted over E-band microwave to achieve large capacity.More validating, Adaptive modulation (AM) schemesconsidered as part of the link path or interference analysistuning. In practice, by using adaptive modulation schemes, weimproved the reliability and performance of a link. In this way,we can ensure that service is available. As depicted in Fig.4,if the E-band link quality degrades, the modulation schemedownshifts step by step, and link BW decreases gradually.Services on E-band links are switched to common-bandlinks before E-band links are completely unavailable. In thiscase, high-priority services within available bandwidth aretransmitted in hitless mode, and packet loss may occur inlow-priority services. Providing to maximize throughput, weenabled AM to provide elastic capacity and also used QoS toguarantee key services. Then, we define certain radio types,configurations, used specific channels, and set the modulationand traffic parameters according to Adaptive Threshold:

Adaptive Threshold(dBm) = Hysteresis(dB)+ Interference(dB)+ CINR(dB)

In wireless communication systems, the performance androbustness are often determined by the signal-to-noise ratio(SNR) from radio link budget:

SNR = Pt +Gt +Gr − Pl − FM − Lsys −Arain (4)

Where Pt is the transmitter (TX) power, Gt and Gr arethe antenna gains at TX and receiver (RX), PL and FM

1024

Q

512Q

256Q

128Q

64Q

32Q

16Q

QPS

K

16Q

32Q

64Q

128Q

256Q

512Q

1024

Q

64Q

32Q

16Q

8Q

QPS

K

QPS

K

8Q16Q

32Q

64Q

Adaptive modulationQoS

E-Band

Common Band

Services

Fig. 4. Adaptive modulation and QoS

denote path loss and fading margin in channel, the system lossLsys include noise power N0 = 10log10(kTB+NF ), WhereT = 290k and noise figure NF = 6dB and implementationloss (IL) at TX and RX, and Arain is rain attenuations inradio link.

As transmission power is restricted and path loss in E-band is very high, the antenna gain becomes very important inguaranteeing long-range point-to-point (PtP) radio links. In thefollowing, the required combined TX-RX antenna gain (sumof antenna gains at TX and RX) is being determined based onlink budget shown in Table 1, where the SNR is chosen as 10dB for providing sufficient power margin in system design.

TABLE II. E-BAND AND COMMON-BAND RADIO LINK BUDGET

E-Band Radio Link Budget Common-Band Radio Link Budget

E-band (71-86)GHz Common-band (17-18)GHzSNR 10 dB SNR 10dBAntenna Gain 49.1 dB Antenna Gain 38.3dBTx Power 10 dB Tx Power 12dBN0 -52dBm N0 -61.5dBmPathloss 92.5 Pathloss 92.5

+20log(83.5GHz) +20log(83.5GHz)+20log[d(km)] +20log[d(km)]

Channel BW 1000M Channel BW 56MModulation 64QAM Modulation 1024QAMAM Enabled AM EnabledQoS Enabled QoS EnabledFade Margin 20dB Fade Margin 20dBConfiguration 1+0 Configuration 4+0Rain Model P530 Rain Model P530R 19.97mm/h R 19.97mm/hAvailability 99.9−99.99 Availability 99.9−99.999

Our implemented CDBLA solution, which bonds the com-mon frequency band (18GHz) together with E-Band (71-86GHz) in a unique manner, offering increased throughput viaCDBLA technique and advanced AM and QoS mechanisms toincrease customer satisfaction and End-users benefit from thefast and stable 6Gbps transmission link. In our implementation,when the full capacity is not available due to high rainfall, AMand QoS will maintain the link and transmit high priority traf-fic. The CDBLA solution combines the long-distance benefitof common frequencies together with the substantial capacityadvantage of the E-Band, ensuring efficient multi-Giga trans-mission. The schema of the proposed CDBLA is shown inFig.5. In each node, we use intra-board radio link bonding totailored and provide superior performance for the particularmicrowave transport solution concern. As a result, in ourimplementation, Link 1-2 and Link 3-4 in Fig.5 configured asan aggregated with XPIC[20] technology. In XPIC technique,


---------------------------------------------------------------------------- 227 ----------------------------------------------------------------------------

a single RF channel utilized for two parallel wireless links,by means of parallel transmission on vertical and horizontalpolarizations independently. Aggregated microwave radio linksis the recommended approach. This scenario does not requiresophisticated devices. XPIC engine consistently gets directfeedback about the wireless interface status and is also capableof adopting load balancing to the unstable wireless link condi-tion (e.g. frequent modulation downshifting). However, XPICaggregation on microwave radios often has the limited numberof parallel links (usually two parallel links), lacks configurationflexibility and may bring some challenges to integration intothe existing network. Though, in this experiment, we ran linkaggregation on external switches to allow a higher number ofparallel aggregated links (allows up to 4 parallel connections.With this XPIC aggregation, we reached to 1.7Gbps throughputthat does not satisfy 5G small-cell requirements. So, we need toadd E-band link to mentioned XPIC aggregation group as ourcomplimentary solution. As a result, all links 1 to 5 groupedas CDBLA solution to pass full 6Gbps Ethernet traffic on air.In this scenario, the switching board collects Ethernet servicebandwidth of each microwave link and allocates traffic basedon the real-time Ethernet bandwidth over each member ofmicrowave aggregation group to achieve almost the same band-width utilization on member microwave links. Moreover, weadded load sharing regardless of Ethernet frame type or length,or whether they provide the same Ethernet bandwidth. Resultof this action provide us with more flexible configuration withmore throughput.

Link 1

Link 5

Link 2

Link 3

Link 4

XPIC enabled

XPIC enabledC

ross Dual B

and Link

Aggregation

E-Band

Common Band

Base Station

Link A

ggregation

E-Band

Common Band

Base Station

Fig. 5. Schema of Implemented CDBLA

III. SIMULATION RESULTS

In this Experiment, Mentum Ellipse, microwave link andmobile backhaul solution engineering software for dimension-ing, planning and optimizing is used to support the creationof accurate link budgets, atmospheric attenuation, selectivefrequency fading. The Fig.5 Shows simulated the microwavelink aggregation with the distance of 3.2km. The designedparameters are simulated according to in TableII. Moreover,Fig.6 shows the simulation ITU recommendation values forour experienced location.

Fig.6 includes longitude and latitude value of CDBLAlocation. Water Concentration, Temperature value, Rain Rate,dN 1% is the refractivity gradient in the lowest 65 m of theatmosphere not exceeded for 1% of an average year and dN50% displays the refractivity gradient in the lowest 65 m of theatmosphere not exceeded for 50% of an average year. MedianRain Height meters above sea level (AMSL) provided in rec-ommendation ITU-R P.839[21], sea level surface refractivityvalue N0 Wet, and Roughness value that obtained from the

Fig. 6. ITU values of CDBLA Simulation

geoclimatic factor K and fading data.

Antenna Height

E-Band Simulation LinkBuilding Hight

Ground Level

Fig. 7. Schema of Microwave link Simulation

In this simulation, nodes are placed based on a clear Lineof Sight (LOS) report, exact on-site coordinates and up todate clutter heights of buildings (Fig.7) that is located ina dense urban area. Fig. 8 are shown for radio parametersto reach 6270 Mbps simulated throughput and 5956 Mbpsreal throughput. In this configuration 5 slots are used as a4+0 with XPIC enable for CBM plus 1+0 for E-Band thataggregated together to provide us with enough capacity for5G small-cells. Other parameters are set according to TableII.

Fig.8 is shown for radio parameters to reach 6270 Mbpssimulated throughput and 5956 Mbps real throughput. In thisconfiguration, five slots are used as a 4+0 with XPIC enable forCBM plus 1+0 for E-Band that aggregated together to provideus with enough capacity. Antenna model, Radio configuration(Tx, Modulation, Frequency Channels, etc.) and environmentalparameters are set based on TableII, which achieved optimumSLA according to Fig.3. Simulation results are shown inTableIII. In this table, Rain (Annual/Worst Month) displays theunavailability due to rain value, total (Annual/Worst Month)presents the total unavailability value in annual against worstmonth value and % Time (Annual/Worst Month) displays theunavailability value in annual against worst month value inpercentage of the time. BER 10-3, BER 10-6 displays theBit Error Rates to calculate quality and availability. In thissimulation availability of CBM and EBM is 99.999% and 99.9,respectively.


---------------------------------------------------------------------------- 228 ----------------------------------------------------------------------------

Fig. 8. Microwave Radio Simulation Parameters

TABLE III. SIMULATED COMMON BAND AND E-BAND MICROWAVE

LINK AVAILABILITY

Slots 4+0 Configuration 1+0 Configuration

Channel - 18 GHz - 56 MHz (Slot No1 - Slot No4)

Rain (Av. % (Std)) Total (Av. % (Std)) % Time

Annual / Worst Month



Channel - 18 GHz - 56 MHz _XPIC (Slot No1 - Slot No4)

BER 10-3 - 435Mbps-1024QAM AB: 99.999918 / 99.99920595

AB: 99.99991791 / 99.99920527

AB: 99.99991791 / 99.99920527

BER 10-6 - 435Mbps-1024QAM AB: 99.99986056 / 99.99873976

AB: 99.99986042 / 99.99873869

AB: 99.99986042 / 99.99873869

Channel - Eband_1000MHz _V (Slot No5)

BER 10-3 - 4530Mbps-64QAM AB: 99.98211882 / 99.91401364

AB: 99.98211858 / 99.91401183

AB: 99.98211858 / 99.91401183

BER 10-6 - 4530Mbps-64QAM AB: 99.97968978 / 99.90393692

AB: 99.97968947 / 99.90393456

AB: 99.97968947 / 99.90393456

IV. EXPERIMENTAL RESULTS

In Fig.9 we displayed our three traffic aggregated linksfor CDBLA, which is implemented on a dense frequencyurban 5G small-cell network based on the plan (Fig.1). Theimplementation contains two of 0.6m, 18GHz Dual PolarizedAntenna and one 0.6m E-band Single Polarized Antenna oneach side plus Indoor units. The throughput results of CDBLAextracted for sample 10 days shown in Fig10. It depicted thatduring the traffic peak time when the majority of Internet usersare online at the same time, we transferred Ethernet trafficbetween 5.9∼6.1 Mbps that matched with our simulationresults in 19.97mm/h rain rate. Besides, the influence of AMis quite shown, without AM function microwave link droppedcompletely along with bad weather issues which needs on-sitemaintenance and much more outage duration time. However,after enabling the AM function, it tries to mitigate the adverseeffect of rain attenuation during climate change. With AMenabling, There is a trade-off among modulation size, BERvalue and throughput. So, we faced with modulation leveldegradation during fading issues by detecting severe climatesituation through analyzing of related alarms. So adaptivemodulation enables to guarantee the performance of the trafficand shorten recovery times.Fig.11,12 show our previous experiments on CBM and EBMlinks, respectively. Those represent, although the CBM linkhas stable 1.8 Gbps throughput, it couldn’t satisfy the 5Grequirements. And for EBM link, even though we reach tothe 3.9Gbps throughput, it does not combat to severe weatherand couldn’t provide a stable 5G Backhaul solution either.

V. CONCLUSION

In the 5G era, traffic to be backhauled increasesexponentially, posing significant challenges on 5G backhaulnetwork. So, to meet this broadband backhaul requirements,CDBLA approach introduced and analyzed, based on theexponential growth of traffic data rate. Following ourimplementation, the practical records provide us with 6.1

E-band MW

E-band MWAgg Link 1-2

Agg Link 3-4

Indoor Unit

Indoor Unit

Agg Link 1-2

Agg Link 3-4

Laboratory Simulation

Fig. 9. Schema of Microwave link Implementation

6049 6038 6064 6073

0

1000

2000

3000

4000

5000

6000

7000

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101

105

109

113

117

121

125

129

133

137

141

145

149

153

157

161

165

169

173

177

181

185

189

193

197

201

205

209

213

217

221

225

Thro

ughp

ut (M

bps)

Time Samples(Hourly)

Rate of good full-frame bits transmitted (Mbps)Downtime-

Troubleshooting

AdaptiveModulation Enabled

Idle Time

Traffic Peak Time based on User Behaviour Stable Throughput

Fig. 10. Real throughput measurement of CDBLA implementation

1775 1798 1790 1793 1792

200

400

600

800

1000

1200

1400

1600

1800

2000

Thro

ughp

ut (M

bps)

Time Samples (Hourly)

Common Band Radio Throughput Results

Fig. 11. Common Band 4+0 Configuration Throughput Results-Previousexperiment, 1024QAM, BW=56M

3947 3871 3897

36883827

0

500

1000

1500

2000

2500

3000

3500

4000

4500

Thro

ughp

ut (M

bps)

Time Samples-Hourly

E-Band Link Throughput Results

AM Disabled - Outage due to severe weather

AM Enabled

Fig. 12. E-band Band 1+0 Configuration Throughput Results-Previousexperiment, 64QAM, BW=1000M

Gbps data rate that satisfies simulation results. As a result,


---------------------------------------------------------------------------- 229 ----------------------------------------------------------------------------

CDBLA could be a perfect solution for 5G small-cellsnetworks in case of throughput, availability, troubleshootingand distance between small-cells to base stations. Thecomparison of Backhaul technologies presented in TableIV.In this comparison, CDBLA is a optimal (low cost with shorttime troubleshooting) solution for high capacity microwavelinks at 5G network.

TABLE IV. COMPARISON OF BACKHAUL TECHNOLOGIES FOR 5GNETWORKS

5G CBM E-Band Optical CDBLASolution FiberBandwidth Small Medium Large Large

ImplementationLow Low High Low

Cost

ImplementationShort Short Long Short

Period

TroubleshootingShort Short Long Short

Time

DistanceLong Short Long Medium

<100km <3km >10km 2−5km

REFERENCES

[1] J. Shi, L. Lv, Q. Ni, H. Pervaiz, and C. Paoloni, “Modeling andanalysis of point-to-multipoint millimeter wave backhaul networks,”IEEE Transactions on Wireless Communications, vol. 18, pp. 268–285,Jan 2019.

[2] J. Chen, X. Ge, and Q. Ni, “Coverage and handoff analysis of 5g fractalsmall cell networks,” IEEE Transactions on Wireless Communications,vol. 18, pp. 1263–1276, Feb 2019.

[3] X. Ge, H. Jia, Y. Zhong, Y. Xiao, Y. Li, and B. Vucetic, “Energy efficientoptimization of wireless-powered 5g full duplex cellular networks: Amean field game approach,” IEEE Transactions on Green Communica-tions and Networking, vol. 3, pp. 455–467, June 2019.

[4] M. Agiwal, A. Roy, and N. Saxena, “Next generation 5g wirelessnetworks: A comprehensive survey,” IEEE Communications SurveysTutorials, vol. 18, pp. 1617–1655, thirdquarter 2016.

[5] Y. A. Abohamra, M. R. Soleymani, and Y. R. Shayan, “Using beam-forming for dense frequency reuse in 5g,” IEEE Access, vol. 7,pp. 9181–9190, 2019.

[6] Y. Okumura, S. Suyama, J. Mashino, and K. Muraoka, “Recent activitiesof 5g experimental trials on massive mimo technologies and 5g systemtrials toward new services creation,” IEICE Transactions, vol. 102-B,pp. 1352–1362, 2019.

[7] W. Roh, J. Seol, J. Park, B. Lee, J. Lee, Y. Kim, J. Cho, K. Cheun, andF. Aryanfar, “Millimeter-wave beamforming as an enabling technologyfor 5g cellular communications: theoretical feasibility and prototype re-sults,” IEEE Communications Magazine, vol. 52, pp. 106–113, February2014.

[8] T. S. Rappaport, W. Roh, and K. Cheun, “Mobile’s millimeter-wavemakeover,” IEEE Spectrum, vol. 51, pp. 34–58, Sep. 2014.

[9] A. V. Alejos, M. G. Sanchez, and I. Cuinas, “Measurement and analysisof propagation mechanisms at 40 ghz: Viability of site shielding forcedby obstacles,” IEEE Transactions on Vehicular Technology, vol. 57,pp. 3369–3380, Nov 2008.

[10] Y. Zhong, X. Ge, H. H. Yang, T. Han, and Q. Li, “Traffic matchingin 5g ultra-dense networks,” IEEE Communications Magazine, vol. 56,pp. 100–105, August 2018.

[11] R. Yin, Y. Zhang, F. Dong, A. Wang, and C. Yuen, “Energy efficiencyoptimization in lte-u based small cell networks,” IEEE Transactions onVehicular Technology, vol. 68, pp. 1963–1967, Feb 2019.

[12] X. Ge, L. Pan, Q. Li, G. Mao, and S. Tu, “Multipath cooperative com-munications networks for augmented and virtual reality transmission,”IEEE Transactions on Multimedia, vol. 19, pp. 2345–2358, Oct 2017.

[13] A. Hilario-Tacuri and A. Tamo, “Ber performance of mm-wave basedsystems in rainfall scenarios,” in 2018 IEEE XXV International Con-ference on Electronics, Electrical Engineering and Computing (INTER-CON), pp. 1–4, Aug 2018.

[14] [online] Available: https://www.itu.int/rec/R REC-P.676-11-201609-I.

[15] [online] Available: https://www.itu.int/rec/R REC-P.530-17-201712-I/en.

[16] J. Antes and I. Kallfass, “Performance estimation for broadband multi-gigabit millimeter- and sub-millimeter-wave wireless communicationlinks,” IEEE Transactions on Microwave Theory and Techniques,vol. 63, pp. 3288–3299, Oct 2015.

[17] X. Zhao, Q. Wang, S. Geng, Y. Zhang, J. Zhang, and J. Li, “Path lossmodification and multi-user capacity analysis by dynamic rain modelsfor 5g radio communications in millimetre waves,” IET Communica-tions, vol. 13, no. 10, pp. 1488–1496, 2019.

[18] I.-R. P.939-2, “Specific attenuation model for rain for use in prediction,”2005.

[19] A. Hilt, “Microwave hop-length and availability targets for the 5gmobile backhaul,” in 2019 42nd International Conference on Telecom-munications and Signal Processing (TSP), pp. 187–190, July 2019.

[20] Xiangnan Liu, Ming Chen, Peiwen Jiang, and Jingwen Wang, “Researchand implementation of high data rate full digital xpic technique,” in2016 15th International Conference on Optical Communications andNetworks (ICOCN), pp. 1–3, Sep. 2016.

[21] https://www.itu.int/rec/R REC-P.839/enITU-R P.839, “Specific attenua-tion model for rain for use in prediction,” 2005.


---------------------------------------------------------------------------- 230 ----------------------------------------------------------------------------

experiment on high capacity backhaul transmission link ...transmission link aggregation solution for...

Documents